fitting curve to a stack of time series

I have a stack of signals A(mxn), where m is the number of signals(lets say 20) and n is the length of each signal (lets say 500). I want to fit a smooth curve within the stack.therefore It is good to Imagine the data as a point cloud of values in a 2D plane where for each n value there are 24 m values.I do not have any label values to assign to signals therefore I wonder if it is feasible at all to use a parametric model in such a scenario for fitting a curve? I would appreciate any hint or link to relative materials Best